Hi Jason
I did install gfortran, according to the manual
apt-get install csh flex gfortran g++ xorg-dev zlib1g-dev libbz2-dev
plus the packages listed as required on your site
http://jswails.wikidot.com/ubuntu-linux#toc6
Serial amber passed all the tests. I then compiled the parallel version with cuda support, after running the configure_openmpi script.
This installed ok, but 44/52 tests failed. The diff file contained the same message for each test. Here is an example.
---------------------------------------
possible FAILURE: check mdout.dif
/usr/local/amber12/test/cuda/nmropt/pme/temp
57c57
< t = 0., dt = 0.00100, vlimit = -1.00000
> t = 0., dt = 0.00100, vlimit = 20.00000
### Maximum absolute error in matching lines = 2.10e+01 at line 57 field 9
### Maximum relative error in matching lines = 2.10e+01 at line 57 field 9
---------------------------------------
The numbers are identical for all the tests (except line and field numbers). Is this anything to worry about?
Thanks
Paul
-----Original Message-----
From: Jason Swails [mailto:jason.swails.gmail.com]
Sent: 08 November 2012 12:57
To: AMBER Mailing List
Subject: Re: [AMBER] compiler errors
On Thu, Nov 8, 2012 at 7:32 AM, Brown, Paul <P.E.Brown.warwick.ac.uk> wrote:
> Dear Jason
> Thanks for your reply. I was trying to get it to work on an old system
> running OSX 10.4. I've given up on this now as Macports wouldn't
> install the required dependencies, and installed on Ubuntu 12.10
> instead. It worked great following your instructions, though configure
> initially failed as the linker was unable to find -lgfortran. Adding
> the following link fixed this
>
> ln -s libgfortran.so.3 /usr/lib/x86_64-linux-gnu/libgfortran.so
>
I would be wary of doing this. I should probably try Ubuntu 12.10 on a VM and update my directions accordingly. Please run the test suite to make sure that everything still works as expected. My guess is that libgfortran.so was not present because some package should have been installed that wasn't (did you install gfortran?)
All the best,
Jason
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Nov 09 2012 - 04:30:03 PST
